What can realistically be automated in quoting?
Quoting consists of several stages and not all require a human. A machine handles repetitive work well: turning a request description into specific catalog items, applying the customer-group discount by rules, calculating margin, checking availability and generating a PDF.
A human stays where judgment is needed: deciding on a substitute for an unavailable item, negotiating terms, approving the offer before sending. Good automation is this split: the system prepares a draft, the rep approves or fixes it. This cuts response time without losing control.
Where to start — how to prepare the catalog?
Automation is only as good as the data it works on. The foundation is a tidy catalog: names, manufacturer indexes, categories and technical parameters (e.g. DN/PN for valves, cross-section for cables, grade for steel). The more canonical parameters, the better the item matching from description.
Practical first steps:
- Import manufacturer price lists (often PDF) into one database — with indexes and purchase prices.
- Unify categories and add canonical parameters for the most important product groups.
- Write down discount rules: which customer groups, what discount on manufacturer/category/name.
- Define a per-product pricing mode — discount or margin — and default margins for “on request” items.
How does matching items from natural language work?
The heart of automation is turning a description (“90-degree elbow dia 50 stainless, 20 pcs”) into a specific catalog item. It’s done in two stages: first a candidate prefilter by fuzzy matching (resilient to inflection, typos and diacritics, with a bonus for a hit manufacturer index), then choosing the best match.
As a result the customer doesn’t need to know catalog symbols, yet the system still hits the right item — even when the request arrived as a plain email. Prices are calculated in code (discount, margin), not “guessed” — important, because the automation must never get the price wrong.
How to automate “on request” items (the RFQ loop)?
Some items have no fixed catalog price — it must be obtained from the manufacturer. This is the most “drift-prone” stage in a manual process: the rep sends emails, waits, forgets who they wrote to. Automation closes it in an RFQ loop: the system sends a price request, the manufacturer enters the purchase price via a simple portal, and the system applies margin and inserts the ready price into the offer.
That’s how OferIQ works: a request from email or form → a draft offer with matched items → an RFQ loop for “on request” items → a ready PDF with a case number. All under the rep’s control, who approves before sending. If you distribute in, say, electrical and automation (/en/industries/elektryka-automatyka/), see the product (/en/product/) and the spreadsheet comparison (/en/compare/oferiq-vs-excel/).